indeed. it's interesting, it takes 1h10m to spool 512G, and then 2h to write it from spool to tapes. so it could easily manage to prepare the stream for tapes on the fly, maybe using a buffer of some gigs.
another thing, it spools data until the spoolsize reached, afterwards writes the data to tapes, but during the write to tape does't spool further, instead when despooling is done it starts again the spooling. Why could it either prepare the stream on the fly, or if spooling enabled at least do spooling/despooling in the same time, in a pipe like style to reduce the times...? On Monday, December 5, 2016 at 6:01:18 PM UTC+1, Doug Rand wrote: > On 12/05/16 06:50, Robert N wrote: > > Hi, > > Hello. > > > I also have performance issues. Does the spooling make any difference > > by copy jobs? I mean the data is anyway already on the disk, so > > copying over to a spool ( disk ) and then to tape improves > > performance? > > I thought exactly the same thing, and it turned out that I was wrong. > > > what do you mean by "I'm not interleaving my backups on disk anymore" > > and how do you achieve it with parallel jobs? > > I'm doing virtual full backups to LTO-6 tape, which is not quite > identical to copy jobs, but really very close. (Virtual full backups > have to touch more source backups than copy jobs.) > > And I still spool the virtual full backups before writing to tape which > allows me to get 160 MB/s to our LTO-6 tape drive. When I did virtual > full backups directly to tape I only got between 80 MB/s and 100 MB/s. > And over the 200 virtual full jobs that I run each week they are from > wall clock perspective faster by spooling to disk and then writing to tape. > > I strongly suggest that you turn both Spool Attributes and Spool Data on. -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. For more options, visit https://groups.google.com/d/optout.
